lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  PHC 
Open Source and information security mailing list archives
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date:	Fri, 08 Aug 2014 09:30:05 -0700
From:	Guenter Roeck <>
To:, Matt Helsley <>
CC:	Jean Delvare <>,
	Benjamin Herrenschmidt <>,
	LKML <>
Subject: Re: [PATCH 5/5] Export the temperatures via hwmon

On 08/08/2014 07:54 AM, Goffredo Baroncelli wrote:
> On 08/07/2014 11:19 PM, Matt Helsley wrote:
>> On Thu, Aug 07, 2014 at 07:50:00PM +0200, Goffredo Baroncelli wrote:
>>> On 08/07/2014 09:36 AM, Guenter Roeck wrote:
>>>> On 08/06/2014 11:52 PM, Jean Delvare wrote:
>>>>> Hi Guenter,
>>>>> On Wed, 06 Aug 2014 23:20:32 -0700, Guenter Roeck wrote:
>>>>>> Patch 4/5 is "Return the fan speed via sysfs: /sys/devices/temperature/fan_level".
>>>>>> So you are saying that returning the fan speed with a non-hwmon attribute works,
>>>>>> but returning it with a hwmon attribute doesn't ? Not really sure if I understand
>>>>>> your logic. Either fan_level doesn't return the fan speed (or an abstraction of it),
>>>>>> or something in your line of argument is inconsistent.
>>>>> fan_level is a fan speed _control_ value, like pwm1. It is not a fan
>>>>> speed monitoring value.
>>>> Ah, ok. The patch description doesn't seem to match, though.
>>>> And why not export it as pwm1, if that is what it is ?
>>>> Guenter
>>> the exported fan_level value is a coefficient near proportional to the speed [*];
>>> so it is not the speed nor the pwm.
>>> I tried to read the pwm/speed value, but when I did it, every 5/6 seconds the
>>> fan seemed to stop for 1s, then the speed raised.... So I stopped the test.
>>> These patches (the first two) solved a real issue: with the last kernels this
>>> driver doesn't work at all, and the fan go to maximum speed (very loud !)
>>> The other three are an improvement.
>>> When (if) these patches will be accepted I want to write another solution,
>>> but definitely not now. And even if it would work for me, it is very likely
>>> that will  be accepted because nobody is able to test it on all hardware.
>>> BR
>>> G.Baroncelli
>>> [*] It is a bit more complicated: basically the adm1030 controls the fan speed
>>> on the basis of an external temperature sensor (the "case" sensor).
>>> Higher is this temperature higher is the fan speed.
>>> The pwm of the fan is computed on the basis of the following value:
>>> 	- min_temp
>>> 	- range_temp -> max_temp=min_temp+range_temp
>>> 	- min_pwm_value
>>> and of course
>>> 	- "case" sensor
>>> The fan_level is an index computed on the basis of the "cpu" temperature (
>>> which is different from the "case" temperature referred above). This
>>> temperature is used to update the min_temp above.
>>> So the fan speed id computed on the basis of two external sensor:
>>> - cpu sensor
>>> - case sensor.
>>> To complicate further the things, the range_temp is set to an undocumented
>>> value: this parameter is set via a register (0x25, bit 2:0), which accepted
>>> values between 0x00 and 0x04. However it is set to 0x07
>>> See
>>> and the adm1030 datasheet.
>>> I am reluctant to change this thing because this driver is for an obsolete
>>> platform (apple stopped the powermac in 2004/2005), and bad or good I have to
>>> suppose that it works. Moreover I am not in the position to test the changes
>>> outside my hardware (1 powermac DP@...z mdd)
>> This seems like a valuable explanation of the meaning of the fan_level
>> values. When I read the code I was wondering what the magic number "11" was
>> doing for instance and this helps clear it up a little. It's a nit perhaps
>> but some of this information might be useful in the commit message for that
>> patch, the code, or both.
> I will take in account your suggestion. In case of another revision, I will improve the comment.
I just noticed - hwmon_device_register_with_groups() returns ERR_PTR
on errors, not NULL, so checking for a NULL return value won't help
much if there is an error.


To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
More majordomo info at
Please read the FAQ at

Powered by blists - more mailing lists